Low-frequency noise from large wind turbines Henrik Møller a) and Christian Sejer Pedersen Section of Acoustics, Aalborg University, Fredrik Bajers Vej 7-B5, DK-9220 Aalborg Ø, Denmark (Received 5 July 2010; accepted 20 December 2010) As wind turbines get larger, worries have emerged that the turbine noise would move down in fre- quency and that the low-frequency noise would cause annoyance for the neighbors. The noise emis- sion from 48 wind turbines with nominal electric power up to 3.6 MW is analyzed and discussed. The relative amount of low-frequency noise is higher for large turbines (2.3–3.6 MW) than for small turbines (2 MW), and the difference is statistically significant. The difference can also be expressed as a downward shift of the spectrum of approximately one-third of an octave. A further shift of similar size is suggested for future turbines in the 10-MW range. Due to the air absorption, the higher low-frequency content becomes even more pronounced, when sound pressure levels in relevant neighbor distances are considered. Even when A-weighted levels are considered, a sub- stantial part of the noise is at low frequencies, and for several of the investigated large turbines, the one-third-octave band with the highest level is at or below 250 Hz. It is thus beyond any doubt that the low-frequency part of the spectrum plays an important role in the noise at the neighbors. V C 2011 Acoustical Society of America. [DOI: 10.1121/1.3543957] PACS number(s): 43.50.Rq, 43.28.Hr, 43.50.Cb, 43.50.Sr [ADP] Pages: 3727–3744 I. INTRODUCTION Wind turbines get larger and larger, and worries have emerged that the noise emitted by the turbines would conse- quently move down in frequency and that the content of low-frequency and infrasonic noise would increase and reach a level, where it may be annoying for the neighbors. The daily press frequently reports on rumbling and annoying noise from large wind turbines, and it is often claimed that it propagates quite far. However, the scientific literature on infrasonic and low-frequency noise from large wind turbines is more limited. A. Low-frequency sound and infrasound A few introductory words about low-frequency sound and infrasound are appropriate. For a more comprehensive review of human hearing at low and infrasonic frequencies, see, e.g., Ref. 1. It is usually understood that the lower limit of the human hearing is around 20 Hz, and the terms infrasound and infrasonic are used with frequencies below this fre- quency. The frequency range 20–200 Hz denotes the low- frequency range (sometimes with a slightly different upper limit). However, as a surprise to many people, the hearing does not stop at 20 Hz. If the level is sufficiently high, humans can hear infrasound at least down to 1 or 2 Hz. The sound is perceived through the ears, but the subjective quality differs from that of sound at higher frequencies. Below 20 Hz, the tonal sensation disappears, the sound becomes discontinuous in character, and a sensation of pressure at the eardrums occurs. At a few hertz, the sensation turns into discontinuous separate puffs, and it is possible to follow and count the sin- gle cycles of a tone. At low and particularly infrasonic frequencies, the loud- ness increases more steeply above the hearing threshold than at higher frequencies, 2–5 and a sound moderately above threshold may be perceived not only loud but also annoy- ing. 6–9 Since there is a natural spread in hearing thresholds, a sound that is inaudible or soft to some people may be loud and annoying to others. Low-frequency noise above the hear- ing threshold may also affect task performance 10 and cause sleep disturbances. 11 There is no reliable evidence of physio- logical or psychological effects from infrasound or low-fre- quency sound below the hearing threshold (see, e.g., Ref. 12). Infrasound is measured with the G-weighting curve, 13 which covers the frequency range 1–20 Hz. At the normal hearing threshold for pure tones, 2,8,14–17 the G-weighted level is in the order of 95–100 dB. G-weighted sound pres- sure levels below 90 dB 13 or 85 dB 18 are normally not con- sidered to be detectable by humans. B. Previous studies Many studies deal theoretically with generating mecha- nisms of low-frequency noise in wind turbines, whereas origi- nal information on low-frequency noise from complete wind turbines is more limited. In the following, only horizontal-axis turbines are considered. Hubbard and Shepherd 19,20 reviewed the literature on wind turbine noise especially emphasizing studies carried out at NASA for more than two decades and comprising tur- bines up to 4.2 MW. It was observed and explained by nu- merical models that harmonics of the blade-passage frequency arise from differences in the inflow wind velocity across the rotor area and, for turbines with the rotor down- wind of the tower, from impulses created by the passage of the blades through the wake of the tower. In particular, the a) Author to whom correspondence should be addressed. Electronic mail: [email protected]J. Acoust. Soc. Am. 129 (6), June 2011 V C 2011 Acoustical Society of America 3727 0001-4966/2011/129(6)/3727/18/$30.00 Author's complimentary copy
18
Embed
Low-frequency noise from large wind turbines · PDF fileLow-frequency noise from large wind turbines Henrik Møllera) and Christian Sejer Pedersen Section of Acoustics, Aalborg University,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Low-frequency noise from large wind turbines
Henrik Møllera) and Christian Sejer PedersenSection of Acoustics, Aalborg University, Fredrik Bajers Vej 7-B5, DK-9220 Aalborg Ø, Denmark
(Received 5 July 2010; accepted 20 December 2010)
As wind turbines get larger, worries have emerged that the turbine noise would move down in fre-
quency and that the low-frequency noise would cause annoyance for the neighbors. The noise emis-
sion from 48 wind turbines with nominal electric power up to 3.6 MW is analyzed and discussed.
The relative amount of low-frequency noise is higher for large turbines (2.3–3.6 MW) than for
small turbines (� 2 MW), and the difference is statistically significant. The difference can also be
expressed as a downward shift of the spectrum of approximately one-third of an octave. A further
shift of similar size is suggested for future turbines in the 10-MW range. Due to the air absorption,
the higher low-frequency content becomes even more pronounced, when sound pressure levels in
relevant neighbor distances are considered. Even when A-weighted levels are considered, a sub-
stantial part of the noise is at low frequencies, and for several of the investigated large turbines, the
one-third-octave band with the highest level is at or below 250 Hz. It is thus beyond any doubt that
the low-frequency part of the spectrum plays an important role in the noise at the neighbors.VC 2011 Acoustical Society of America. [DOI: 10.1121/1.3543957]
The tone analyses show that tones generally vary in
level and frequency with wind speed. Figure 6 shows tonal
audibility for the most prominent tones of turbines 1–4.
FIG. 4. (Color online) Normalized A-weighted apparent sound power levels
in one-third-octave bands, means of two groups of turbines: � 2 MW and
> 2 MW. Error bars indicate 61 standard error of mean.
FIG. 5. Normalized A-weighted apparent sound power levels in one-third-
octave bands, mean of 36 turbines � 2 MW (bold line) and 9 individual tur-
bines > 2 MW.
FIG. 6. (Color online) Tonal audibility, DLta, as a function of wind speed
for turbines 1–4, reference direction (turbine color code as in Fig. 5).
J. Acoust. Soc. Am., Vol. 129, No. 6, June 2011 H. Møller and C. S. Pedersen: Low-frequency wind-turbine noise 3733
Au
tho
r's
com
plim
enta
ry c
op
y
Values are below 3–4 dB, except for turbine 3 at high
wind speeds. For turbines 1 and 3, the data apply to a tone
that varies with wind speed around 110–145 Hz, approxi-
mately the same frequency range for both turbines. For tur-
bine 2, the data apply to a tone with a nearly constant
frequency around 40 Hz. Turbine 4 has several tones at higher
frequencies, and those in the frequency range 800–1400 Hz
alternately dominate, depending on wind speed. One-third-
octave-band peaks can be identified in Fig. 5 for the two tur-
bines with tonality above 0 dB at 8 m/s (turbine 2, 40 Hz;
turbine 3, 160 Hz).
ISO 1996–2 (Ref. 56) specifies a tone penalty to be
used, when the tonal audibility exceeds 4 dB. National crite-
ria for tone penalty may vary, e.g., Danish regulation
requires that the tonal audibility exceeds 6.5 dB, before a
penalty is given.57
Only one turbine exceeds the 4 dB limit and only at
high wind speeds, where noise regulation may not apply. It
is quite surprising that not even the most distinct tone in the
one-third-octave-band spectra, the 40-Hz tone of turbine 2,
results in a tone penalty. This is most likely an effect of the
critical band used for tone assessment being very wide at
low frequencies. It is outside the scope of the present article
to evaluate if the tones will be perceived as being tonal
despite the lack of tone penalty.
4. Directivity
Figure 7 shows the directivity of the three turbines
measured.
The data differ somewhat between turbines, and it is dif-
ficult to find a general pattern. Both higher and lower levels
are seen in other directions than the reference. At the lowest
frequencies, a low directivity would be expected, but this is
not seen in the data. A measured directivity may reflect a
true directivity, but if the main noise source is at one side in
the rotor plane, e.g., at the down going blade as shown by
Oerlemans and Schepers58 and Oerlemans et al.59 the
measurement in this side is closer to the source, and a false
indication of directivity may result.
A possibly source of error for the directivity data is that
the measurements for the various directions do not always
refer to the same period. Each of the other directions was in
fact measured together with the reference direction, but they
were not all measured at the same time. Only one data set
exists for the reference direction, and thus this cannot apply
to all directions. At low frequencies, poor signal-to-noise ra-
tio may be responsible for large uncertainty.
The direction from the turbine to neighbors is typically
more horizontal than the direction to the measurement posi-
tions. In particular, if sound is radiated from synchronous
vibrations in blades and/or tower, chances are that the radia-
tion will be more perpendicular to the rotor plane and/or the
tower, i.e., close to the horizontal plane. More knowledge is
called for on this issue.
5. Effect of wind speed
Figure 8 shows LWA as a function of wind speed for the
four turbines, where data is available.
The noise increases with wind speed but levels out or
even decreases above 7–8 m/s. The four turbines are all
pitch-controlled, and the observation is in line with the
reports by, e.g., Lee et al.36 and Jung et al.37 for pitch-controlled
turbines.
B. Outdoor sound pressure levels at neighbors
For each of the large turbines, the distance needed for
the A-weighted sound pressure level to decrease to 35 dB
was derived. Pedersen and Waye60 have shown that around
this sound pressure level, the percentage of highly annoyed
persons increases above 5%, and the percentage of annoyed
persons increases above 10% (Pedersen et al.61). Pedersen
and Nielsen62 recommended a minimum distance to neigh-
bors so that the wind turbine noise would be below 33–38
dB. A limit of 35 dB is used for wind turbines, e.g., in Swe-
den for quiet areas.63 Thus, 35 dB seems as a very reasona-
ble limit for wind turbine noise. It is also the limit that
FIG. 7. (Color online) Directivity of turbines 1–3. Wind speed is 8 m/s
except for turbine 2, front, which was measured at 10 m/s (and compared to
reference direction at 10 m/s). Data missing for turbine 2 front at 5 kHz due
to electric noise in the measurement (turbine color code as in Fig. 5).
FIG. 8. (Color online) A-weighted apparent sound power level, LWA, as a
function of wind speed for turbines 1–4 (turbine color code as in Fig. 5).
3734 J. Acoust. Soc. Am., Vol. 129, No. 6, June 2011 H. Møller and C. S. Pedersen: Low-frequency wind-turbine noise
Au
tho
r's
com
plim
enta
ry c
op
y
applies in Denmark in open residential areas (night) and rec-
reational areas (evening, night, and weekend) for industrial
noise64 (but not for wind turbine noise49).
Table I shows the distances for the individual turbines
as well as various key figures at the 35-dB distances.
The minimum distance, where a 35-dB limit is complied
with, varies considerably between the large turbines, even
when the turbines are relatively equal in size (2.3–3.6 MW).
The distance varies from slightly over 600 m to more than
1200 m.
The one-third-octave-band spectra at the same distances
are shown in Fig. 9.
At these distances, the air absorption plays a role. It
affects mainly the high frequencies, and the result is that the
shift of the spectrum towards lower frequencies becomes
even more pronounced than for the source spectrum (com-
pare with Fig. 5).
It is important to note that, for several turbines, the high-
est level for a one-third-octave-band is at 250 Hz or lower,
even when A-weighted levels are regarded (Fig. 9). It is thus
beyond any doubt that the low-frequency part of the spec-
trum plays an important role in the noise at the neighbors
and that the low-frequency sound must be treated seriously
in the assessment of noise from large turbines.
In many cases, A-weighted outdoor levels in excess of
35 dB are allowed. As an example, for houses outside offi-
cial residential or recreational areas, Danish regulation
allows 44 dB.49 For visual reasons, the Danish regulation
has a setback distance for dwellings of four times the total
turbine height, and at this distance, the level is often below
44 dB for a single turbine. However, 44 dB may certainly
occur further away than four times the turbine height, when
there are several turbines together in wind farms. Table II
lists distances to small wind farms, where the A-weighted
sound pressure level is 44 dB, as well as various key figures
at those distances.
C. Sound insulation
During the measurements, there were severe problems
with background noise at the three lowest frequencies.
Eighteen measurements with a signal-to-noise ratio below
1.3 dB were discarded. Consequently, seven room/fre-
quency combinations had to be derived from measurements
in only two or three 3D corners. Two room/frequency com-
binations with measurements from only one 3D corner were
not calculated. Figure 10 shows the sound insulation for the
ten rooms.
For the frequencies 63–200 Hz, with few exceptions,
the rooms have 10–20 dB sound insulation. Toward lower
frequencies, the insulation decreases, while the variation
between rooms becomes larger. Some rooms show very lit-
tle or even negative insulation at certain frequencies. A
single room has unusually high insulation in the 16–31.5
Hz range. This room was a small room used for storage of
furniture and other goods. The room is thus not considered
a typical living room, and its data are discarded in further
calculations.
Be aware that, for each one-third-octave band, the
indoor level refers to the maximum level that people would
normally be exposed to in the room (Sec. II D). Thus, in par-
ticular, for the higher end of the frequency range, the insula-
tion data are lower than traditional insulation data employed
for technical purposes, where room average levels are typi-
cally used.
1. Shortcomings of insulation measurements
A shortcoming with the measurement method used is
that the exposure is focused at the facade of the house. In the
situation of the house being exposed to noise from wind tur-
bines, the whole house, including the roof and, at low fre-
quencies, also the back of the house, will be exposed to
nearly the same sound. In the measurement situation, these
other surfaces receive much less sound due to loudspeaker
directivity, higher distance to the loudspeaker, shadowing,
etc.
TABLE I. Key figures at the distances from a single turbine, where the total A-weighted sound pressure level is 35 dB. Distances are given as slant distance
to rotor center, which, for actual turbine heights, is close to horizontal distance.
it might have been possible to increase the signal level by
measuring one one-third-octave band at a time rather than
the whole low-frequency range simultaneously.
D. Indoor sound pressure levels at neighbors
Figure 11 shows indoor one-third-octave-band levels for
all 81 combinations of 9 turbines and 9 rooms at the distance
with a total A-weighted outdoor sound pressure level of 35
dB. Be aware that the indoor levels estimate the maximum
level that people would normally be exposed to in the room
and not the average level of the room (Sec. II D).
Large differences are seen between turbine/room combi-
nations. Most of the variance is attributed to differences in
the room sound insulation, except at 63 and 80 Hz, where
both room and turbine contribute equally. Values in the
upper end of the range at 40 Hz are due to high emission
from a single turbine, whereas high values at 200 Hz are due
to low sound insulation of a single room.
It is seen from the inserted hearing threshold (dashed
line), that the low-frequency sound will be audible in many
turbine/room combinations, mainly at the highest of the low
frequencies. The sound will not be very loud, but as men-
tioned in the introduction, low-frequency sound can be
annoying only slightly above the hearing threshold (Sec. I
A), and some people may be annoyed by the sound.
Figure 12 shows indoor levels for the situations from
Table II where the A-weighted outdoor sound pressure level
from a wind farm is 44 dB.
Here, there will be audible sound somewhere in all
rooms and with all turbines. In more than half of the cases
(48 out of 81), the normal hearing threshold is exceeded by
more than 15 dB in one or more one-third-octave bands, and
there is a risk that a substantial part of the residents will be
annoyed by the sound.
For continuous noise, to avoid sleep disturbance, WHO
recommends an indoor limit of 30 dB for the A-weighted
sound pressure level,65 but also notes that, if the noise
includes a large proportion of low-frequency noise, “a still
TABLE II. Key figures at the distances where the total A-weighted sound pressure level is 44 dB. Wind farm with two rows of each six identical turbines, 300
m distance between turbines in both directions (200 m for small turbines). Observer point centered at long side. Distances are given as slant distance to closest
FIG. 11. Indoor A-weighted one-third-octave-band sound pressure levels at
the distance from a single turbine, where the total A-weighted outdoor sound
pressure level is 35 dB (see Table I); 81 turbine/room combinations. Dashed
line is hearing threshold according to ISO 389–7 (Ref. 28) (colors indicate
the turbine, color code as in Fig. 5).
3736 J. Acoust. Soc. Am., Vol. 129, No. 6, June 2011 H. Møller and C. S. Pedersen: Low-frequency wind-turbine noise
Au
tho
r's
com
plim
enta
ry c
op
y
lower guideline value is recommended, because low-fre-quency noise . … can disturb rest and sleep even at lowsound pressure levels.” How much lower is not stated, but
unless the level above 200 Hz is exceptionally low, the total
A-weighted sound pressure level will obviously exceed, e.g.,
25 dB in many of the cases in Fig. 12.
1. Danish indoor limit
The Danish indoor evening/night limit for LpALF in
dwellings of 20 dB (Ref. 18) does not apply to measurements
in single positions but to levels measured by the method
mentioned in Sec. II D. The method uses the power average
of measurements in three positions: one position near a cor-
ner of the room and two positions where the complainant
perceives the noise as being loudest. Assuming that the com-
plainant appoints such positions adequately, the result of the
entire method—the power average with a corner position—
will still be a level close to the maximum.
It is not possible to find the maximum LpALF by simply
adding the one-third-octave-band levels from Fig. 11 or Fig.
12, since the various one-third-octave bands may have their
maximum in different areas of the room. However, 40 of
the 81 turbine/room combinations of Fig. 12 exceed an
A-weighted level of 20 dB for at least one one-third-octave
band in the 10–160 Hz frequency range, and it is reasonable
to believe that the total for that frequency range, LpALF, will
exceed 20 dB for even more combinations.
It should be mentioned that wind turbines have been
exempt from the general Danish guidelines for low-fre-
quency sound since 2006, when the regulation for wind tur-
bines was updated.49 The argument was that indoor LpALF
will not exceed 20 dB, if the normal outdoor limits are com-
plied with.66 This may be true for smaller turbines, but as
seen, the indoor level may easily exceed 20 dB with large
turbines above 2 MW.
IV. GENERAL DISCUSSIONS
A. Noise versus turbine size
The data material gives a useful overview of the sound
power emitted from wind turbines of different sizes, and,
with caution, it may be possible to use the data to estimate
the apparent sound power level of future, larger turbines.
Figure 13 repeats the data for LWA from Fig. 1, now with an
extrapolation toward higher nominal electric power, and
data for the regression line inserted.
The regression line in Fig. 13 corresponds to the follow-
ing connection between the apparent sound power, PA, and
the nominal electric power, PE:
PA ¼ constant1 � PE=1MWð Þslope=10dB(2)
where slope is the slope of the regression line, and con-stant1 can be derived from the last term of the regression
line. Since the slope is 11.0 dB, the exponent is 1.10, mean-
ing that the apparent sound power increases more than pro-
portionally to the nominal electric power. Thus, to the
extent that turbines follow the trend of the regression line, a
turbine of double size emits more than the double sound
power.
The area A of the circle, within which a certain noise
limit is exceeded, is of particular interest. The radius of the
circle can be found by solving Eq. (1) with respect to d, and,
if omitting the atmospheric absorption, which mainly has
effect at high frequencies and at long distances, it is found
that the area is proportional to the apparent sound power. Af-
ter insertion of Eq. (2), it follows that
A¼ constant2 �PA
¼ constant2 � constant1 �PE
1MW
� �slope=10dB
(3)
where constant2 depends on the noise limit.
FIG. 13. (Color online) Apparent sound power level (LWA) as a function of
The differences (3.6 and 2.2 dB) are in the same order
of magnitude as the differences in the present investigation
(compare with Fig. 4).
A comparison with data of the present investigation con-
verted to octave bands shows very similar values in the two
investigations, see Fig. 15. Data from the two investigations
for the same power group are not significantly different at
any frequency. (There is no overlap in original data.)
D. Tonal components
Søndergaard and Madsen70 conclude (1) that the
“frequency spectra of the aerodynamic noise from the rotorblades of the largest wind turbines does not deviate signifi-cantly from the spectra for smaller wind turbines. Thismeans that for the aerodynamic noise the low frequencyrange is not more prominent for large turbines than for smallturbines,” (2) that the observed “slightly higher . … relativeamount of low frequency noise . … is mainly caused by geartones at frequencies below 200 Hz,” and (3) that this “is notunusual for prototypes and usually the fully developed com-mercial wind turbines are improved on the noise emission,especially concerning audible tones in the noise.”
However, these conclusions are not substantiated by
adequate statistics or other data analyses. The separation of
aerodynamic noise and gear noise referred to is not
explained, and data are not given. Regarding the develop-
ment of noise from prototypes to commercial turbines, no
FIG. 14. (Color online) Normalized A-weighted apparent sound power lev-
els in octave bands, means for two groups of turbines: < 2 and � 2 MW.
Data from van den Berg et al.,69 Appendix D. Error bars indicate 61 stand-
ard error of mean. (None of the large turbines was measured in the 31.5-Hz
octave band).
3738 J. Acoust. Soc. Am., Vol. 129, No. 6, June 2011 H. Møller and C. S. Pedersen: Low-frequency wind-turbine noise
Au
tho
r's
com
plim
enta
ry c
op
y
data or references are given. If the turbines of the present
project are considered, it is unclear, whether turbines 5–11
are prototypes or not, since the turbines are anonymous, and
the informations diverge between reports. The original
report43 only specifies turbines 1–4 as prototypes, but a sum-
marizing report70 refers to all the turbines above 2 MW as
prototypes. If turbines 5–11 are indeed prototypes, this
means that the third conclusion is made without data for
large commercial turbines. If, on the other hand, turbines
5–11 are commercial turbines, it is worth noting that some
of these also have obvious one-third-octave-band peaks (Fig.
5), and that their noise emissions (LWA or LWALF) are not
lower than those of turbines 1–4, perhaps on the contrary
(Fig. 1).
Regarding reduction of tonal noise, Søndergaard and
Madsen refer to the tone penalty as a means to guarantee
that the tones are actually reduced, before the turbines are
put on the market, and they use expressions like “the neces-sary tone reduction”70 and “… reduced to a level wherethere is no penalty according to Danish rules….”43,70 They
have evidently ignored that the results of their tone analyses
will not release a tone penalty to any of the turbines (Sec.
III A 3).
A closer look at the data reveals that, even when some
of the one-third-octave-band peaks at low frequencies are
very distinct, the peaks are not in general responsible for the
difference between small and large turbines. Figure 16
shows an imagined situation, where all peaks below 200 Hz
have been removed from the large turbines by replacing the
level at the peaks with levels obtained by linear interpolation
between the levels in the two adjacent one-third-octave
bands. One to three peaks have been removed for each tur-
bine, except for turbine 4, which does not have peaks in this
frequency range. Only removal of the 40-Hz peak of turbine
2 affects the mean of the large turbines by more than 1.0 dB.
Generally, the large turbines are still above the mean of
the small turbines in the low-frequency range. The difference
between the means of large (> 2 MW) and small turbines
(� 2 MW) is still significant in the same one-third-octave
bands as they were with the peaks [63–160 Hz (unchanged
23.6, 17.0, 19.2, 18.9), one-sided p = (0.003, <0.001, 0.006,
0.005, 0.003)].
The striking similarity with the spectra from van den
Berg et al.69 (Fig. 15) supports that the spectra for the large
turbines from the present project, including the tones, are
representative for wind turbines of such size.
E. Ground reflection
In the calculations of sound pressure levels at the neigh-
bors, the ground reflection is accounted for by adding 1.5 dB
to the direct sound. As mentioned in Sec. IIC, the 1.5-dB
value is used by Danish regulation.49 Swedish guidelines
add 3 dB to the direct sound (for distances up to 1000 m),71
a value that also follows from ISO 9613–2 (Ref. 47) for the
lowest octave-frequency band mentioned, 63 Hz, irrespec-
tive of the ground surface. During measurements of sound
emission from the turbines,46 it is assumed that the ground
reflection adds as much as 6 dB to the direct sound. Cer-
tainly, a reflecting board is used under the microphone, but
the board has only little effect at low frequencies, where the
assumed 6-dB reflection is due mainly to the ground itself.
Possible destructive interference between the direct
sound and the ground reflection due to elevation of the re-
ceiver above ground will have little impact at low frequen-
cies. For example, for a source height of 75 m, a horizontal
distance of 800 m, and a receiver height of 1.5 m, the delay
between the direct sound and the ground reflection will only
be 0.8 ms, which corresponds to a first dip in the sound trans-
mission at 625 Hz.
On this background, it is reasonable to suspect that the
addition of 1.5 dB for the ground reflection is too low at low
frequencies, and that higher values up to a theoretical maxi-
mum of 6 dB would be more appropriate. Thus, the procedure
used to calculate outdoor sound pressure levels at the neigh-
bors is likely to underestimate the low-frequency sound.
FIG. 16. Normalized A-weighted apparent sound power levels in one-third-
octave bands, individual turbines > 2 MW and mean of 36 turbines � 2
MW. Peaks in one-third-octave bands below 200 Hz have been removed
from the large turbines by replacing the levels at the peaks by levels
obtained by linear interpolation between the levels at the two adjacent one-
third-octave-band frequencies (turbine color code as in Fig. 5).
FIG. 15. (Color online) Normalized A-weighted apparent sound power lev-
els in octave bands, means for two groups of turbines: < 2 and � 2 MW and
from two investigations: van den Berg et al. (Ref. 69), Appendix D and pres-
ent investigation (converted to octave bands).
J. Acoust. Soc. Am., Vol. 129, No. 6, June 2011 H. Møller and C. S. Pedersen: Low-frequency wind-turbine noise 3739
Au
tho
r's
com
plim
enta
ry c
op
y
F. Windows
The measurements of sound insulation were made with
closed windows. However, in large parts of the world, many
people prefer to sleep with the windows at least slightly
open, and WHO recommends that noise limits should permit
this.65,72 In Denmark, indoor measurements of low-fre-
quency noise are usually made with closed windows, but if
the complainant finds that the noise is louder with open win-
dows, measurements should also be made for this situation.18
Therefore, it would have been appropriate to measure the
insulation also with slightly open windows and to estimate
the resulting indoor sound pressure levels accordingly.
G. Estimated sound power spectra for even largerturbines
In Sec. III A 2, the spectral difference between small
and large turbines was seen in terms of differences in the
normalized apparent sound power levels for certain one-
third-octave bands. As an alternative way, Fig. 17 shows the
mean normalized spectra of large and small turbines, but
with the data for small turbines shifted one third of an octave
down in frequency.
The two curves are very close in the main frequency
range, meaning that the spectrum has maintained its shape
but shifted about one third of an octave down in frequency
from the small to the large turbines (compare with Fig. 4).
Differences at the lowest frequencies may be real or be the
result of uncertainty due to high background noise at these
frequencies, a matter that is not fully expounded in the data
material.
For the reader who might think that a shift of a single
third octave is very modest, it is worth noting that it is the
same as the musical interval of a major third, nearly the dif-
ference between two adjacent strings on a guitar.
The logarithmic means of the nominal electric power of
the small and large turbines are around 650 kW and 2.6
MW, respectively, thus the downward spectral shift of
approximately one third of an octave relates to an upward
shift of the nominal electric power by a factor in the order of
4. It would thus be appropriate to suggest a further down-
ward spectral shift of the same amount for future turbines in
the 10-MW range.
As a supplement to the linear regression and the extrap-
olation for LWA in Fig. 13, estimated spectra have been con-
structed for turbines around 2.5, 5, and 10 MW for possible
(and cautious) use in future projects. Figure 18 shows a
sixth-order polynomial regression of the relative spectrum
for the turbines of the present project above 2 MW.
Table III gives relative one-third-octave-band levels for
2.5 MW turbines from the regression and, for 5 and 10 MW
turbines, data shifted one sixth and one third of an octave,
respectively, down in frequency. In addition, the table gives
estimated absolute levels based on the linear regression of
LWA in Fig. 13. Note that the estimates are based on means
of turbines and that they do not include a safety margin as
mentioned in Sec. IV B.
The table values for the absolute level in one-third-
octave bands are shown in Fig. 19.
H. Atmospheric conditions
All previous calculations assume spherical sound propa-
gation, i.e., a 6 dB reduction of sound pressure level per dou-
bling of distance. During certain atmospheric conditions,
e.g., with temperature inversion or low-level jets, there may
be a sound reflecting layer in a certain height, and thus the
propagation beyond a certain distance is more like cylindri-
cal propagation, which only gives 3 dB reduction per dou-
bling of distance. This was observed for low frequencies,
e.g., by Hubbard and Shepherd19 and explained, e.g., by Zor-
umski and Willshire73 and Johansson.74 Above sea, Swedish
guidelines generally assume cylindrical propagation beyond
a distance of 200 m,71 a distance supported by data by Bolin
et al.,75 who showed reflection in a height in the order of
100–200 m.
With cylindrical propagation beyond 200 m, the follow-
ing equation applies (for distances above 200 m):
FIG. 17. (Color online) Normalized apparent sound power levels in one-
third-octave bands. Mean of two groups of turbines: � 2 and > 2 MW,
group of turbines � 2 MW shifted one third of an octave down in frequency.
(Turbine 6 disregarded above 2 kHz, see Sec. III A 2.)
FIG. 18. (Color online) Sixth-order polynomial regression (bold line) for
mean of normalized apparent sound power levels (dots and thin line) for the
turbines > 2 MW (Turbine 6 disregarded above 2 kHz, see Sec. III A 2.)
3740 J. Acoust. Soc. Am., Vol. 129, No. 6, June 2011 H. Møller and C. S. Pedersen: Low-frequency wind-turbine noise
Au
tho
r's
com
plim
enta
ry c
op
y
Lp¼LWA�20 dB � log10
200 m
1m
� ��10 dB � log10
d
200 m
� �
�11 dB�a �dþ1:5 dB: (4)
Table IV and Fig. 20 show key figures and sound pressure
levels in one-third-octave bands, respectively, at the distan-
ces from the turbines, where the A-weighted sound pressure
level has decreased to 35 dB, assuming cylindrical propaga-
tion beyond 200 m.
Much longer distances (1414–3482 m) are needed than
with pure spherical propagation, and the low-frequency char-
acter of the spectrum has become even more pronounced
(compare with Table I and Fig. 9). Cylindrical propagation
may thus explain case stories, where rumbling of wind tur-
bines is claimed to be audible kilometers away. A worst-case
scenario combining temperature inversion with a wind park
acting as a line source in a certain distance range could theo-
retically reduce the geometrical attenuation in that range to
zero. However, more knowledge is needed about atmos-
pheric conditions and the occurrence of various phenomena.
Also other phenomena related to the atmospheric condi-
tions deserve some attention. It is normally assumed that the
TABLE III. Estimated relative and absolute A-weighted sound power levels for turbines around 2.5, 5, and 10 MW based on sixth-order polynomial approxi-
mation of mean relative spectrum for turbines above 2 MW from Fig. 18 and LWA from linear regression of Fig. 13. Relative levels moved, respectively, 1/6
and 1/3 of an octave down for 5 and 10 MW turbines. Approximation adjusted by þ0.38 dB to achieve a total relative spectrum of 0 dB, which the mean of rel-
ative data (and its approximation) does not necessarily sum up to. Note that the estimates are based on means of turbines and that they do not include a safety
margin as mentioned in Sec. IV B.
Relative to LWA Absolute
Frequency (Hz) 1=3-octave-band levels Octave-band levels 1=3-octave-band levels Octave-band levels
FIG. 19. (Color online) Estimated A-weighted sound power levels in one-
third-octave bands for turbines around 2.5, 5, and 10 MW. Values and
assumptions are taken from Table III.
J. Acoust. Soc. Am., Vol. 129, No. 6, June 2011 H. Møller and C. S. Pedersen: Low-frequency wind-turbine noise 3741
Au
tho
r's
com
plim
enta
ry c
op
y
wind speed increases logarithmically with increasing height
above ground, starting from zero speed at a height equal to
the roughness length of the ground surface. Thus, knowing
the roughness length, the wind speed at all heights can be
determined from measurements in a single height. The wind
speed in a height of 10 m is used as a reference for measure-
ments of wind turbine noise.46
However, several studies have shown that actual wind-
speed profiles vary a lot and often deviate substantially from
the assumed logarithmical profile.76–79 In a stable atmos-
phere, which often exists at night, variations with height can
be much larger than assumed with high wind speed at turbine
height and little wind at ground. A large variation of wind
speed across the rotor area increases the modulation of the
turbine noise, and the normal “swish–swish” sound turns
into a more annoying, “thumping,” impulsive sound as
reported by, e.g., van den Berg27,80,81 and Palmer.82 The
effect is more prominent with large wind turbines, where the
difference in wind speed between rotor top and bottom can
be substantial. The effect is usually not reflected in noise
measurements, which are mainly carried out in the daytime,
when the logarithmic profile is more common.
Another consequence of large wind speed variation with
height is that the turbine may emit noise corresponding to a
high wind speed—and much higher than assumed from the
wind speed measured at 10 m—while it is all quiet at the
ground. Thus, there is more turbine noise than expected and
less wind; hence, the turbine noise will not be masked with
natural wind-induced sound, as it might have been with the
assumed logarithmic wind profile.
Several authors have argued that the logarithmic wind-
speed profile and the 10-m reference height are inadequate
with the size of modern turbines (e.g., Refs. 77, 78, 80, 83),
and a revised IEC 61400-11 will use the actual wind speed
in the turbine hub height as a reference.84 Wind profiles and
statistics for the actual place can then be applied in noise
prediction and regulation.
V. CONCLUSIONS
The results confirm the hypothesis that the spectrum of
wind turbine noise moves down in frequency with increasing
turbine size. The relative amount of emitted low-frequency
noise is higher for large turbines (2.3–3.6 MW) than for
small turbines (� 2 MW). The difference is statistically sig-
nificant for one-third-octave bands in the frequency range
63–250 Hz. The difference can also be expressed as a down-
ward shift of the spectrum of approximately one third of an
octave. A further shift of similar size is suggested for tur-
bines in the 10-MW range.
When outdoor sound pressure levels in relevant neigh-
bor distances are considered, the higher low-frequency con-
tent becomes even more pronounced. This is due to the air
absorption, which reduces the higher frequencies a lot more
than the lower frequencies. Even when A-weighted levels
are considered, a substantial part of the noise is at low fre-
quencies, and for several of the investigated large turbines,
the one-third-octave band with the highest level is at or
below 250 Hz. It is thus beyond any doubt that the low-fre-
quency part of the spectrum plays an important role in the
noise at the neighbors.
Indoor levels of low-frequency noise in neighbor distan-
ces vary with turbine, sound insulation of the room, and
position in the room. If the noise from the investigated large
turbines has an outdoor A-weighted sound pressure level of
44 dB (the maximum of the Danish regulation for wind tur-
bines), there is a risk that a substantial part of the residents
will be annoyed by low-frequency noise even indoors. The
Danish evening/night limit of 20 dB for the A-weighted
noise in the 10–160 Hz range, which applies to industrial
noise (but not to wind turbine noise), will be exceeded some-
where in many living rooms at the neighbors that are near
the 44 dB outdoor limit. Problems are much reduced with an
outdoor limit of 35 dB.
The turbines do emit infrasound (sound below 20 Hz),
but levels are low when human sensitivity to these frequencies
is accounted for. Even close to the turbines, the infrasonic
FIG. 20. A-weighted sound pressure levels in one-third-octave bands at
the distances, where the total A-weighted sound pressure level is 35 dB (see
Table IV). Cylindrical propagation assumed from 200 m (turbine color code
as in Fig. 5).
TABLE IV. Key figures at the distances, where the total A-weighted sound pressure level is 35 dB, cylindrical propagation assumed beyond 200 m. Distances
are given as slant distance to rotor center, which, for actual turbine heights, is close to horizontal distance.
3742 J. Acoust. Soc. Am., Vol. 129, No. 6, June 2011 H. Møller and C. S. Pedersen: Low-frequency wind-turbine noise
Au
tho
r's
com
plim
enta
ry c
op
y
sound pressure level is much below the normal hearing thresh-
old, and infrasound is thus not considered as a problem with
turbines of the investigated size and construction.
The low-frequency noise from several of the investigated
large turbines comprises tones, presumably from the gearbox,
which result in peaks in the corresponding one-third-octave
bands. The tone penalty does not guarantee that the tones are
removed or reduced, since they are not sufficiently distinct to
release a penalty at all. The spectral difference between large
and small turbines remains statistically significant, even if the
one-third-octave-band peaks are removed.
The above conclusions are based on data for turbines in
the range of 2.3–3.6 MW nominal electric power. It must be
anticipated that the problems with low-frequency noise will
increase with even larger turbines.
The emitted A-weighted sound power increases propor-
tionally to the nominal electric power or likely even more.
Consequently, large turbines affect the same area—or possi-
bly even larger areas—with noise, when compared to small
turbines with the same total installed electric power.
There are differences of several decibels between the
noise emitted from different turbines of similar size, even for
turbines of the same make and model. It is therefore not fea-
sible to make calculations down to fractions of a decibel and
believe that this holds for the turbines actually set up. A
safety margin must be incorporated at the planning stage in
order to guarantee that the actual erected turbines will com-
ply with noise limits. An international technical specification
exists for this, but it is often not used.
Under certain atmospheric conditions, e.g., temperature
inversion, the noise may be more annoying and—in particu-
lar the low-frequency part—propagate much further than
usually assumed. More knowledge is needed on such phe-
nomena and their occurrences.
ACKNOWLEDGMENTS
The measurements were carried out by Delta. Financial
support was obtained from the Energy Research Programme
under the Danish Energy Agency, and from Aalborg
University.
1H. Møller and C. S. Pedersen, “Human hearing at low frequencies,” Noise
Health 6(23), 37–57 (2004).2L. D. Whittle, S. J. Collins, and D. W. Robinson, “The audibility of low
frequency sounds,” J. Sound Vib. 21(4), 431–448 (1972).3H. Møller and J. Andresen, “Loudness of pure tones at low and infrasonic
frequencies,” J. Low Freq. Noise Vib. 3(2), 78–87 (1984).4M. A. Bellmann, V. Mellert, C. Reckhardt, and H. Remmers, “Perception
of sound and vibration at low frequencies,” collected papers from the JointMeeting “Berlin 99” of ASA, EAA and DAGA, Berlin, Germany (1999),
Abstract in J. Acoust. Soc. Am., 105, 1297.5ISO 226 Acoustics—Normal Equal-Loudness-Level Contours (Interna-
tional Organization for Standardization, Geneva, Switzerland, 2003).6J. Andresen and H. Møller, “Equal annoyance contours for infrasonic
frequencies,” J. Low Freq. Noise Vib. 3(3), 1–9 (1984).7H. Møller, “Annoyance of audible infrasound,” J. Low Freq. Noise Vib.
6(1), 1–17 (1987).8Y. Inukai, N. Nakamura, and H. Taya, “Unpleasantness and acceptable
limits of low frequency sound,” J. Low Freq. Noise, Vib., Act. Control,
19(3), 135–140 (2000).9J. K. Subedi, H. Yamaguchi, Y. Matsumoto, and M. Ishiharatil,
“Annoyance of low frequency tones and objective evaluation methods,”
J. Low Freq. Noise, Vib., Act. Control. 24(2), 81–96 (2005).
10K. P. Waye, J. Bengtsson, A. Kjellberg, and S. Benton, “Low frequency noise
‘pollution’ interferes with performance,” Noise Health. 4(13), 33–49 (2001).11K. P. Waye, A. Clow, S. Edwards, F. Hucklebridge, and R. Rylander,
“Effects of nighttime low frequency noise on the cortisol response to
awakening and subjective sleep quality,” Life Sci. 72, 863–875 (2003).12B. Berglund and T. Lindvall (Eds.) “Community noise,” Archives of the
center for Sensory Research, 2 (1), Stockholm University and Karolinska
Institute, 1995, prepared for the World Health Organization.13ISO 7196, Acoustics—Frequency-Weighting Characteristic for Infrasound
Measurements (International Organization for Standardization, Geneva,
Switzerland, 1996).14N. S. Yeowart and M. J. Evans, “Thresholds of audibility for very low-fre-
quency pure tones,” J. Acoust. Soc. Am. 55(4), 814–818 (1974).15S.Yamada, T. Kosaka, K. Bunya, and T. Amemiya, “Hearing of low fre-
quency sound and influence on human body,” in Proceedings of Confer-ence on Low Frequency Noise and Hearing, Aalborg, Denmark (May 7–9,
1980), pp. 95–102.16U. Landstrom, R. Lundstrom, and M. Bystrom, “Exposure to infrasound—
Perception and changes in wakefulness,” J. Low Freq. Noise Vib. 2(1),
1–11 (1983).17T. Watanabe and H. Møller, “Low frequency hearing thresholds in pres-
sure field and in free field,” J. Low Freq. Noise Vib. 9(3), 106–115 (1990).18Lavfrekvent støj, infralyd og vibrationer i eksternt miljø (Low Frequency
Noise, Infrasound and Vibrations in the External Environment), Orientering
fra Miljøstyrelsen, nr. 9 (Danish Environmental Protection Agency, 1997).19H. H. Hubbard and K. P. Shepherd, “Aeroacoustics of large wind
turbines,” J. Acoust. Soc. Am. 89(6), 2495–2508 (1991).20K. P. Shepherd and H. H. Hubbard, “Physical characteristics and percep-
tion of low frequency noise from wind turbines,” Noise Control Eng. J.
36(1), 5–15 (1991).21G. Guidati, R. Bareiß, and S. Wagner, “An investigation of blade-tower-
interaction noise (BTI) for horizontal axis wind turbines in upwind and
downwind configuration. First steps towards modeling of aeroelastic
effects,” in Proceedings of the 8th IEA Joint Action Symposium on Aero-dynamics of Wind Turbines, Lyngby, Denmark (1994), pp. 249–255.
22M. L. Legerton, D. M. J. P. Manley, J. W. Sargent, D. J. Snow, and
P. Styles, “Low frequency noise & vibration levels at a modern wind
farm,” in Proceedings of the 25th International Congress on NoiseControl Engineering, Inter-Noise 96, Liverpool, Turkey, pp. 459–462.
23K. Betke, M. Schultz-von Glahn, O. Goos, and H. Remmers, “Messung der
Infraschallabstrahlung von Windkraftanlagen” (Measurement infrasound emis-
sion from wind turbines), in Proceeding of DEWEK ’96. 3rd German WindPower Conference Wilhelmshaven, Germany (1996), pp. 207–210.
24K. Betke and H. Remmers, “Messung und Bewertung von Tieffrequentem
Schall,” (Measurement and assessment of low-frequency sound), Fortschr.
Akustik, DAGA 1998, pp. 472–473.25J. Jakobsen, “Infrasound emission from wind turbines,” J. Low Freq.
Noise, Vib., Act. Control 24(3), pp. 145–155 (2005).26J. Jakobsen, “Danish guidelines on environmental low frequency noise,
infrasound and vibration,” J. Low Freq. Noise, Vib., Act. Control 20(3),
141–148 (2001).27G. P. van den Berg, “Do wind turbines produce significant low frequency
sound levels,” 11th Int. Meeting on Low. Frequency Noise Vib. its Con-trol., Maastricht, The Netherlands (2004).
28ISO 398-7, Acoustics—Reference Zero for the Calibration of AudiometricEquipment—Part 7: Reference Threshold of Hearing under Free-Fieldand Diffuse Field Listening Conditions (International Organization for
Standardization, Geneva, 2005).29C. S. Pedersen and H. Møller, Vurdering af lavfrekvent støj og infralyd fra
decentrale el-elproducerende anlæg (Assessment of Low-Frequency Noiseand Infrasound from Local Electricity-Producing Plants) (Department of
Acoustics, Aalborg University, Aalborg, Denmark, 2005).30“The measurement of low frequency noise at three UK wind farms,’ Hayes
McKenzie Partnership, Ltd., for Department Trade Industri (DTI), Con-
tract No. W/45/00656/00/00, URN No. 06/1412, UK, 2006.31A. Moorhouse, D. Waddington, and M. Adams, “Proposed criteria for the
assessment of low frequency noise disturbance,” Project Rep. DEFRA
NANR45, University of Salford, 2005.32A. Moorhouse, D. Waddington, and M. Adams, “Procedure for the assess-
ment of low frequency noise complaints,” Project Rep. DEFRA NANR45,
University of Salford, 2005.33J. Jakobsen, Lavfrekvent støj fra vindmøller(Low-Frequency Noise from
J. Acoust. Soc. Am., Vol. 129, No. 6, June 2011 H. Møller and C. S. Pedersen: Low-frequency wind-turbine noise 3743
Au
tho
r's
com
plim
enta
ry c
op
y
34Bekendtgørelse om miljøgodkendelse af hurtigfærgeruter(Statutory Orderon High-Speed Ferries), Bekendtgørelse No. 821(Danish Ministry of the
Environment, København, 1997).35Vurdering af lavfrekvent støj fra færger, part 2: Revideret metode til
beregning af lydtrykniveauet indendørs (Assessment of Low-FrequencyNoise from Ferries, Part 2: Revised Method for Calculating the SoundPressure Level Indoors), Delta Akustik & Vibration, Arbejdsrapport fra
Miljøstyrelsen nr. 10, 1997.36S. Lee, S.-H. Shin, C. Cheong, S.-S. Jung, and W. Cheung, “Low-fre-
quency noise emission characterisation of upwind-type large wind
turbines,” in Proceedings of the International Congress on Noise ControlEngineering, Inter-Noise 2007, Istanbul, Turkey.
37S. S. Jung, W.-S. Cheung, C. Cheong, and S.-H. Shin, “Experimental iden-
tification of acoustic emission characteristics of large wind turbines with
emphasis on infrasound and low-frequency noise,” J. Korean Phys. Soc.
53(4), 1897–1905 (2008).38W. J. Gastmeier and B. Howe, “Recent studies of infrasound from indus-
Akustik & Vib., 1996.52C. Simmons, “Measurement of sound pressure levels at low frequencies in
rooms. Comparison of available methods and standards with respect to
microphone positions,” Acta. Acust. 85(1), 88–100 (1999).53S. Pedersen, H. Møller, and K. P. Waye, “Indoor measurements of noise at
low frequencies—Problems and solutions,” J. Low Freq. Noise, Vib., Act.
Control 26(4), 249–270 (2007).54J. Brunskog and F. Jacobsen, “Measurements of low-frequency noise in
rooms,” Memorandum (Danish Environmental Protection Agency, 2008).55S. Wagner, R. Bareiß, and G. Guidati, Wind Turbine Noise (Springer, Ber-
lin, Germany).56ISO 1996-2Acoustics—Description, Measurement and Assessment of
Environmental Noise—Part 2: Determination of Environmental NoiseLevels, 2nd ed. (International Organization for Standardization, Geneva,
2007).57Maling af ekstern støj fra virksomheder (Measurement of External Indus-
trial Noise), Vejledning nr. 6 (Danish Environmental Protection Agency,
1984).
58S. Oerlemans and G. Schepers, “Prediction of wind turbine noise and com-
parison to experiment,” in Proceedings of Second International Meetingon Wind Turbine Noise, Lyon, France (2007).
59S. Oerlemans, M. Fisher, and T. Maeder, “Reduction of wind turbine noise
using optimized airfoils and trailing-edge serrations,” AIAA. J. 47(6),
1470–1481 (2009).60E. Pedersen and K. P. Waye, “Perception and annoyance due to wind tur-
bine noise—A dose-response relationship,” J. Acoust. Soc. Am. 116(6),
3460–3470 (2004).61E. Pedersen, F. van den Berg, R. Bakker, and J. Bouma, “Response to
noise from modern wind frams in The Netherlands,” J. Acoust. Soc. Am.
126(2), 634–643 (2009).62T. H. Pedersen and K. S. Nielsen, “Genevirkning af støj fra vindmøller,”
(Annoyance noise from wind turbines), Report 150, Delta Akustik & Vib.,
1996.63Buller fran vindkraft—Riktvarden for ljud fran vindkraft (Noise from Wind
Turbines—Recommended Limits For Sound From Wind Turbines), Natur-
vardsverket (Stockholm, 2009).64Ekstern støj fra virksomheder(External Ind. Noise), Vejledning nr. 5 (Dan-
ish Environmental Protection Agency, 1984).65B. Berglund, T. Lindvall, and D. H. Schwela (eds.), Guidelines for Com-
munity Noise (World Health Organization, Geneva, Switzerland, 1999).66“Høring af udkast til bekendtgørelse om støj fra vindmøller” (Hearing of
draft statutory order on noise from wind turbines), Miljøstyrelsen, 30.
October2006.67IEC TS 61400-14, Wind turbines—Part 14: Declaration of Apparent
Sound Power Level and Tonality Values (International Technical Commis-
sion, Geneva, 2005).68S. Broneske, “Comparison of wind turbine manufacturers’ noise data for
use in wind farm assessments,” in Proceedings of Third InternationalMeeting on Wind Turbine Noise, Aalborg, Denmark (2009).
69F. van den Berg, E. Pedersen, J.e Bouma, and R. Bakker,
“WINDFARMperception—Visual and acoustic impact of wind turbine
farms on residents,” Final Rep. 3., University of Groningen, University of
Gothenburg, June 2008.70B. Søndergaard and K. D. Madsen, “Low frequency noise from large wind
turbines—Summary and conclusions on measurements and methods,”
Rep. AV 140/08, Delta (revised version December 2008.71 “Ljud fran vindkraftverk” (Sound from wind turbines), Rapport 6241,
Naturvardsverket, Stockholm, Sweden, 2001.72Night Noise Guidelines for Europe (World Health Organization, Copenha-
gen, Denmark, 2009).73W. E. Zorumski and W. L. Willshire, “Downwind sound propagation in an
atmospheric boundary layer,” AIAA. J. 5, 515–523 (1989).74L. Johansson, “Sound propagation around off-shore wind turbines,” in
Proceedings of 10th International Congress on Sound and Vibration,
Stockholm, Sweden (2003), pp. 1481–1487.75K. Bolin, M. Boue and I. Karasalo, “Long range sound propagation over a
sea surface,” J. Acoust. Soc. Am. 126(5), 2191–2197 (2009).76G. P. van den Berg, “Wind gradient statistics up to 200 m altitude over flat
ground,” in Proceedings of First International Meeting on Wind TurbineNoise, Berlin, Germany (2005).
77P. Botha, “The use of 10 m wind speed measurements in the assessment of
wind farm developments,” in Proceedings of First International Meetingon Wind Turbine Noise, Berlin, Germany (2005).
78W. K. G. Palmer, “Uncloaking the nature of wind turbines—Using the sci-
ence of meteoroly,” in Proceedings of Second International Meeting onWind Turbine Noise, Lyon, France (2007).
79D. Bowdler, “Wind shear and its effect on noise assessment,” in Proceedings ofThird International Meeting on Wind Turbine Noise, Aalborg, Denmark (2009).
80G. P. van den Berg, “Effects of the wind profile at night on wind turbine
sound,” J. Sound Vib. 277, 955–970 (2004).81G. P. van den Berg, “The beat is getting stronger: The effect of atmos-
pheric stability on low frequency modulated sound of wind turbines,” J.
Low Freq. Noise, Vib., Act. Control 24(1), 1–24 (2005).82W. K. G. Palmer, “A new explanation for wind turbine whoosh—wind
shear,” in Proceedings of Third International Meeting on Wind TurbineNoise, Aalborg, Denmark (2009).
83M. Almgren, S. Schonfeld, and J. Gronlund, “Sound emission and sound
propagation for wind turbines in forest terrains,” in Proceedings of ThirdInternational Meeting on Wind Turbine Noise, Aalborg, Denmark (2009).
84B. Søndergaard, “The next version of the IEC 61400-11 measurement
method,” in Proceedings of Third International Meeting on Wind TurbineNoise, Aalborg, Denmark (2009).
3744 J. Acoust. Soc. Am., Vol. 129, No. 6, June 2011 H. Møller and C. S. Pedersen: Low-frequency wind-turbine noise